With domestic inequalities and “populism” taking center stage, the changes in global income inequality have understandably moved out of the focus. My access to the data—many of them still best obtained from the World Bank especially for poorer countries not covered by LIS—has also diminished since I left the World Bank. But one can still put together a quick update (thanks also to the contribution by Christoph Lakner) that covers the period up to 2013. One small technical point is worth making at the outset. When one compares, as I will do here, 2013 results with the past, going back to 1988, one faces the following choice: either (1) to use the best 2000-2013 numbers which are all based on detailed micro data and 100 percentiles of the population from each country, plus better income data from India, or (2) to compress these numbers into country-deciles to make them more comparable with the 1988 data (when indeed we had much less detailed information with many fewer fractiles). I decided here to go with the solution No. 2—but if I were to compare a more recent period only (say, after 2000), I would have preferred to use No. 1.
Do household survey data track reasonably well what we know from National Accounts? The answer is yes. The Figure below shows the average annual global growth rates of GDP per capita and mean income from household surveys over five-year periods, all expressed in 2011 international dollars. The two blue lines move together reaching both the peak growth of about 3% per capita per annum (pc; pa) in the period prior to the Global Financial Crisis before dropping to only 1% pc pa in the next five years.
But the striking and important thing is what we cannot see from National Accounts but can see from household surveys: the dramatic increase in the global median income (income at the 50th global percentile). This positional income which reflects high growth rates of the relatively poor populations in Asia (China, India, Vietnam, Indonesia etc.) has throughout the past 25 years always increased faster than the global mean income, and in the most recent period (2008-13) the gap between the median and mean growth has soared: the median income went up by an average rate of 6% pc pa while the mean income grew by only 1%.
The shrinkage of the distance between the mean and the median is often taken as an indicator of reduced inequality (for asymmetric distributions). And this is indeed the case here. In 1988, the mean per capita income of the world was just over $PPP 4000 and the median just over $PPP 1000; a quarter-century later, these amounts were respectively $PPP 5500 and $PPP 2200. So the mean-to-median ratio has decreased from 4-1 to 2.5-1. The global Gini coefficient went down from 0.69 to 0.62; the global Theil index from 0.92 to 0.73. Most of the decline took place in the last five-year period. By 2008, the global Gini was 0.67, just slightly below its value at the time of the fall of the Berlin Wall. What happened after 2008 was the slowdown or negative growth of rich countries and continued fast growth of Asia, combined with the absence of further within-national increases in income inequality in big countries like China, Russia, UK and Brazil. Even in the United States, income inequality went down as the result of the shock to highest incomes during the crisis, before shooting back again after 2013 (the period not covered here though).
Does this mean that everything is fine? Not really. Measures of inequality such as the mean-to-median ratio or the top 1% share are fragmentary: they take into account only what is happening at two points of the distribution. Synthetic measures like the Gini are, in that sense, better (because they take into account the entire distribution) but they compress all that information into one number. To get a better sense of what is happening we want to look at various parts of the distribution and at various measures.
Let me give two examples. The growth of the median, which is, as we have just seen, a very good and encouraging development, has its other facet: it leaves behind those below the median whose incomes do not grow as fast as the median. If we take, for example, the income level equal to ½ of the median, which is often used as a measure of relative poverty or inequality, then we note that the number of people below that income level has increased by 300 million and the percentage of the global population falling short of ½ of the median has barely changed (see Figure below). It used to be 28% of the world population in 1998, it is 26% now.
The share of income received by the global 1% has also, despite the shrinkage of global inequality, remained unchanged. In 1988, its share was 11.3%; it then increased to around 13.5% in 2003 and 2008 before going back to 11% as the crisis struck the rich economies which “supply” most of the people in the global top 1%. Given that we are probably missing an increasing number of the super-rich or that they are hiding their assets more than in the past, it is very likely that the true share of the top 1% has even increased.
We thus have only apparently paradoxical developments over the past 25 years: on the one hand, strongly rising global median income and the shrinkage of global inequality when measured by the synthetic indicators like the Gini or Theil; but, on the other hand, the rising share of the global top 1% and increasing number of people in relative poverty (mostly in Africa). The last point opens up again the vexed question of lack of convergence of Africa and its growing falling behind Asia (and of course the rest of the world).
So, is the world becoming better, as Bill Gates wants us to believe? Yes, in many ways, it is: the mean income in 2013 is almost 40% higher than in 1988, and global inequality is less. But is there a bad news too? Yes: the same share of the world population is being left behind and the top 1% are getting ever further away and richer than everybody else. So, we have, at the same time, the growth of the global “median” class and an increase in world-wide polarization.